BACKGROUND OF THE INVENTION
I. Field of the Invention
The present invention relates to data flow through a frame relay network and, more particularly, to a system and method for controlling the data flow from a LAN (local area network) station through a frame relay network when the frame relay network becomes congested.
II. Background and Prior Art
As users have moved along the path toward decentralized processing, local connectivity problems have been solved by attaching network elements--workstations, minicomputers, and microcomputers--via local area networks (LANs). Further, as users wanted to have global communications that make any device anywhere appear as if it were attached to the local LAN, LANs spread across the globe have become internetworked across wide area networks (WANs). Because LAN interconnect data is bursty and unpredictable, "fast packet" services, which provide bandwidth on demand (such as frame relay), are the networking technologies of choice for interconnecting LANs.
Frame relay is a multiplexed data networking service supporting connectivity between internetworking units (IWUs), such as bridges and routers, and between IWUs and carrier networking equipment. IWUs which are normally located at the customer premises may be called customer premises equipment (CPE). In frame relay, which can be considered the successor to the CCITT X.25 packet standards, error correction and flow control are handled at network end points, i.e., at the IWUs. Frame relay accelerates the process of routing packets through a series of switches to a remote location by eliminating the need for each switch to cheek each packet it receives for errors before relaying it to the next switch. Instead, only the IWUs at the network end points cheek for errors. This error treatment increases performance and reduces bandwidth requirements, which in turn can reduce communications costs and decrease the number of packet handling devices needed in the network.
Furthermore, flow control is conducted by the IWUs at the end points of the network only--no flow control is presently conducted between the user and the network end point. Thus, the IWU connecting with the frame relay network is charged with controlling the flow of data frames into the network, i.e., slowing down or speeding up the data traffic into the network based upon the network's present load.
The frame relay technologies are defined as having to provide the IWU with a congestion notification so that the IWU can assist in clearing the network by not forwarding any additional data frames until the network is adequately clear for transmission. These congestion notification messages are denoted as FECN (forward explicit congestion notification) and BECN (backward explicit congestion notification) and indicate that the network is congested. These indicators, as well as the rate control concept based upon CIR (committed information rate), are used to protect the frame relay network from congestion. The IWU receiving either of these indicators (FECN or BECN) responds accordingly, as will be discussed below.
With regard to the LAN data traffic, in a normal LAN environment, i.e., in an environment where there is no LAN interconnection, stations connected to the LAN perform certain network control functions, such as error correction and flow control. With regard to flow control, LAN stations conforming to the IEEE 802.2 recommendation will provide a logical link connection (LLC) protocol element which implements a dynamic window flow control mechanism. This mechanism when activated will reduce the window size through which data may pass to a fraction of its original value, and in accordance with the success of subsequent transmissions, the window size will automatically increase, until it returns to its original value.
Due to the use of the IWU at the boundary between the frame relay network and the LAN, however, the frame relay network is transparent to the LAN station and, thus, the LAN station cannot perform any flow control function. The LAN station has no way of knowing that the frame relay network is congested as it does not know that the frame relay network is even there. Thus, the LAN station of the prior art must depend upon the IWU at the network boundary to receive all of the data destined to cross the frame relay network and to control its flow when the network becomes congested.
FIG. 1 shows an example of a network 10 having four local area networks (LAN1 12, LAN2 14, LAN3 16 and LAN4 18) interconnected through a frame relay network 15. Station A1 22 is connected directly to LAN1 12 while station A2 24 is connected directly to LAN4 18. Bridges B1 26 and B2 28 perform bridging functions to interconnect LAN1 12 and LAN2 14 (B1 26) and to interconnect LAN3 16 and LAN4 18 (B2 28). Routers R1 30 and R2 32 perform the routing functions to interconnect LAN2 14 and LAN3 16 to the frame relay network 15, respectively. Routers R1 30 and R2 32 represent the IWUs at the frame relay network end points as discussed above.
In present systems, when the frame relay network 15 becomes congested, it sends to routers R1 30 and R2 32 an indication to slow the flow of data into the network. These are the B ECN and FECN indicators discussed above. Upon receiving either of these indicators, router R1 30 (R2 32) will store the frames of data that it receives from LANs 1 and 2 (LANs 3 and 4) until the data can be safely transmitted over the network 15. In the meantime, however, stations connected to each LAN continue to forward frames of data to router R1 30 (and router R2 32) as if the network were clear. Router R1 30 (R2 32) merely stores this data in a queue in its buffer storage area until there is an indication from the network that a higher volume of data may pass.
For instance, when a frame is sent from a user (which is a router in the above example) to its target through a frame relay network, if the network experiences congestion in the same direction as the frame to going, FECN will be activated. If the congestion is in the opposite direction, BECN will be activated. The frame relay network expects users to slow data flow when BECN or FECN is activated, the manner in which the traffic is slowed is at the user's discretion. Many users, however, do not take appropriate actions and continue to pump frames into the network which results in frames being discarded by the network.
In the cases where a router takes appropriate action and slows the data traffic to the congested network, there is a problem in the case where the router is receiving too much data for transmission across the frame relay network. If the network is congested, the router stems the data flow by storing the data that it receives in internal buffers but, because it is receiving so much data from the LAN stations, it may run out of buffer storage space for the incoming data. Because it has no mechanism for indicating to the attached stations to slow or stop the data flow, this data will be discarded and lost.
Even where the router or other CPE has sufficient storage space for storing all of the outgoing data frames, the router cannot retrieve the stored data frames quickly enough for transmission onto the network after the network is clear as this requires much processing effort. Thus, the router (or other IWU) becomes the bottleneck to the LAN data traffic. This problem is especially common when the router receives high volume, bursty data traffic from the LAN stations for transmission over the frame relay network.
Thus, it can be seen that with present systems, there is no mechanism for efficiently interconnecting local area networks (LANs) through a frame relay network. Present systems do not allow workstations connected to the LANs to provide flow control for data that the workstations are transmitting over the frame relay network when the network becomes congested. Instead, the network end points are solely responsible for this task which ultimately results in the end points themselves becoming the bottlenecks to the network traffic as their buffer storage overflows and data is lost.
SUMMARY OF THE INVENTION
The system and method of the present invention provide a mechanism, for the purpose of efficiently interconnecting local area networks (LANs) across a frame relay network, by which LAN stations provide data flow control when the frame relay network becomes congested. The system of the invention comprises a Station Manager at the network end point internetworking unit (IWU) for managing the IWU, a LAN Manager for managing one or more LANs connected to the end point IWU, and Station Managers at each of the LAN stations managing the respective LAN station. In operation, the IWU receives from the frame relay network a congestion notification and the IWU Station Manager sends this notification to its managed LAN station Station Managers. The LAN station Station Managers each examine the notification against its outstanding data to be transmitted and slows down data traffic that is destined for the congested portion of the frame relay network.
BRIEF DESCRIPTION OF THE DRAWINGS
While the technical description concludes with claims particularly pointing out and distinctly claiming that which is regarded as the invention, details of a preferred embodiment of the invention may be more readily ascertained from the following technical description when read in conjunction with the accompanying drawings, where:
FIG. 1 is a diagram of a representative communications network.
FIG. 2 depicts another communications network in which the present invention is embodied.
FIG. 3 is a simplified diagram illustrating a protocol data unit (PDU) used in a source routing system.
FIG. 4 is a simplified diagram illustrating a PDU used in a transparent tree routing system.
FIG. 5 is a simplified diagram illustrating the fields of a data frame as transmitted across a FR network.
FIG. 6 depicts a diagram illustrating the flow of messages between the frame relay network, a connected IWU (router), and connected LAN stations when the frame relay network is congested and source routing is being used by the LAN stations.
FIG. 7 depicts a diagram illustrating the flow of messages between the frame relay network, a connected IWU (router), and connected LAN stations when the frame relay network is congested and transparent tree routing is being used by the LAN stations.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
A system embodying the present invention is shown in FIG. 2. As can be seen in the figure, the system 40 for interfacing with the frame relay FR network 15 comprises a LAN workstation A1 42, a router R3 44 and a LAN Manager 46. The system 40 further comprises LAN1 and LAN2 12 and 14, respectively, and bridge B1 26. LAN1, LAN2 and bridge B1 remain unchanged in implementing the present invention.
As can be seen, each of the system components, LAN workstation A1 42, router R3 44 and newly added LAN Manager 46, comprises a plurality of elements. In particular, referring first to the LAN Manager 46, the LAN Manager comprises a physical layer element, a media access control (MAC) protocol element, a logical link control (LLC) protocol element, a LAN Manager element and a Station Manager element. Each element performs a different and necessary function of the LAN Manager.
The physical layer element, which is analogous to the physical layer of the seven-layer OSI Model, provides the actual physical connectivity to LAN2 14 by supporting the LAN medium such as unshielded twisted pair, coaxial or fiber optic cabling for the transmission of bit streams across the particular LAN medium.
The MAC element is responsible for managing the traffic on the LAN. It performs functions such as determining when the LAN media is free to transmit data, detects collisions of data on certain types of LANs, and determines when transmission should be re-initiated.
The LLC element provides the interface between the LAN and the user layers. LLC can be configured to provide a very basic service (connectionless) or a very elaborate service of dealing with connection-oriented operations. In addition, an LLC element of a LAN station conforming to IEEE 802.2 provides data flow control in the form of a dynamic flow control window mechanism. This mechanism when activated will reduce the window size through which data may pass to a fraction of its original value, and in accordance with the success of subsequent transmissions, the window size will automatically increase, until it returns to its original value.
The Station Manager element is an element implemented in each LAN station. The Station Manager manages the particular LAN station on which it resides. Further, each Station Manager has a functional address (usually in the form of a MAC address). By way of the functional address, each Station Manager may be addressed.
The LAN Manager element manages all of the Station Managers within its domain, where a domain may spread physically across multiple LANs. The LAN Manager manages the Station Managers by accepting unsolicited reports from them and by issuing commands for them to execute. Like the Station Managers, the LAN Manager also has a functional address and uses the Station Managers functional addresses for communicating therewith.
Referring now to LAN station A1, LAN station A1, similar to the LAN Manager, comprises a physical layer element, a MAC layer element, an LLC layer element and a Station Manager element. These elements act in the same manner and perform the same function as described above for the LAN Manager.
In addition, the LAN station A1 comprises a "higher layers" element. This element is merely a representation of any additional function, above the LLC layer functions, that the workstation may have. Some examples are software development and engineering design application programs.
Referring now to the router, router R3 likewise comprises a Station Manager element, a physical layer element and a MAC protocol element, and, in addition, a frame relay (FR) protocol element, and an "other router functions" element. The physical layer element, like the physical layer element of the LAN Manager, provides the actual physical connectivity to LAN2 14 in the same manner described. In addition, however, it provides physical connectivity to the FR network 15 by supporting either a classical RS422, X.21, V.35 (fast RVX) or High Speed Serial Interface (HSSI).
The MAC layer and Station Manager elements act in the same manner and perform the same functions described above for the LAN Manager.
The FR Protocol element supports functions which allow the router to be attached to the FR network. In addition to interfacing with the physical layer element, the FR Protocol element acts as a data link control layer by supporting I.121 functions such as frame delimiting, frame multiplexing/demultiplexing, frame check sequence, etc.
Furthermore, the FR Protocol element supports the BECN and FECN mechanisms by which the frame relay network notifies its end points that the network is congested. For example, when the network is congested, the network will issue a congestion notification frame to its end points indicating that the network is congested (via the BECN bit or the FECN bit). Further, in that same frame, the network will identify which data link connection identifier (DLCI), or DLCIs, is associated with the congestion. The FR Protocol element receives congested DLCI, performs the necessary mapping (to be discussed further below), and forwards these results to its Station Manager for forwarding to the LAN Manager. The LAN Manager notifies the appropriate LAN Stations that the frame relay network is congested and to slow the data traffic flow to the router.
The "other router functions" element performs such other router functions as offering flow control mechanisms as well as source routing and nonsource routing features (to be discussed below).
The method and system of the present invention operate with LANs using either type of routing: source routing and nonsource (normally, transparent tree) routing. In source routing, the source element (host) dictates the routes of the protocol data unit (PDU) through the internet. The source element places the addresses of the "hops" (the intermediate elements or IWUs) in the PDU. An example of a source routing PDU is shown in FIG. 3. Generally, a source routing PDU comprises a number of fields, including a Control field, a Destination Address (DA) field, a Source Address (SA) field, a Routing Information (RI) field, a Data field, and a Frame Check Sequence (FCS) field for error checking and correction. In the RI field, the source element, such as a LAN station, places the addresses of the IWUs through which the data must pass to reach the destination element. In the figure, these IWUs include the numbers (addresses) of LANa, Bridge a (Ba), Router a (Ra), LANb, Bb, Rb, and LANn, Bn, and Rn.
In nonsource routing, the IWUs make decisions about the route and do not rely on the PDU to contain information about the route. FIG. 4 illustrates an example nonsource routing PDU. As with the source routing PDU, the nonsource routing PDU comprises a Control field, a Data field and an FCS field. In addition, the nonsource routing PDU comprises a Source MAC Address field and a Destination MAC Address field. The IWUs in the network use these two addresses to properly forward the PDU to the destination element through the use of "look-up" tables.
FIG. 5 illustrates the format of a frame relay network (FR) frame. The FR frame comprises flags at either end to indicate the start and end of the frame, a Data field and a Frame Cheek Sequence field for error checking and correction. In addition, the FR Frame comprises an Address field consisting of data link connection identifier (DLCI) (high order) and DLCI (low order) fields, a C/R field, an EA 0 and an EA 1 field, a DE field, and FECN and BECN fields. The C/R (command/response indication), the EA (address field extension bit), and the DE (discard eligibility indicator) fields are not particularly relevant to the invention and will not be discussed. The DLCI (high order) and DLCI (low order) fields are used together to represent the address of the destination element. The FECN and BECN fields, as discussed to a certain extent above, are used by the frame relay network to notify its end points that the network is congested, either in the forward or backward direction.
In general, the system of the present invention operates as follows. The stations on the local area networks (such as A1 on LAN1) transmit data to router R3 for transmission across frame relay network 15. When frame relay network 15 becomes congested, the network 15 sends a congestion notification to router R3 (by setting the BECN bit in the FR frame for a predetermined period of time). The frame relay network includes with the congestion notification frame the DLCI of the congested portion of the network. This indicates to router R3 that the network is congested at that DLCI.
As mentioned, the FR Protocol element maps the congested DLCI to the affected LAN Stations. Where source routing is being used, the FR Protocol element parses the frame and maps the congested DLCI as reported by the frame relay network to determine the affected remote LAN segment. Where nonsource routing is being used, the FR Protocol element determines from the look-up table which LAN stations are affected by the congested DLCI (LAN stations are mapped to a particular DLCI in the look-up table).
The router's Station Manager conveys this information to the LAN Manager which, in turn, conveys the information to the remainder of the Station Managers by sending a message to their respective functional addresses. Each of the Station Managers then determines whether its station is or will be transmitting data to the affected portion of the network. If so, the dynamic window flow control mechanism is used to slow down the data flow to the router and, in particular, the data flow destined for the affected portion. If, after a period of time, the network remains congested at the previous DLCI, the FR Protocol element resends the message to its Station Manager that the network is congested and the process, as discussed above, is repeated.
FIGS. 6 and 7 illustrate message flows between the network and the network components in implementing the present invention, first, where source routing is required (FIG. 6) and, second, where transparent tree routing is used (FIG. 7). Referring to FIG. 6, after the FR network becomes congested, it sends a congestion notification (with an indication of the congested DLCI) to router R3 (which is received by the FR Protocol Element). Along with performing its required functions, i.e., frame delimiting, frame multiplexing/demultiplexing, frame cheek sequence, etc., the FR Protocol Element parses the frame so that the DLCI is in the form of a segment number so that the LAN stations may understand it. The FR Protocol element then forwards this information (in the form of a congestion notification along with the segment number associated with the congested DLCI) to its Station Manager (in router R3). R3 Station Manager then sends this information to its LAN Manager 46. Because the LAN Manager can communicate with each of the Station Managers, it forwards this information to each of them in messages sent to their respective functional addresses. The Station Managers forward this information to their respective LLC protocol elements. The LLC protocol elements examine the segment number associated with the congested DLCI against their outstanding logical link connections (LLCs) which may or may not be to the congested segment. If there is a match, the corresponding LLC protocol element will slow down the data traffic associated with that segment number.
As was discussed, the workstations use the known window flow control mechanism for slowing down the traffic to that segment. Using this mechanism, the workstation's data output is slowed in accordance with a "window" size. The window size is reduced where there is network congestion to a particular segment as identified by the FR Protocol element. The window size is slowly increased (and, thus, data traffic), based upon successful transmissions of data frames. If the network remains congested at the particular DLCI, the FR Protocol element is notified of the congestion by a BECN or FECN flag and the process discussed above is repeated so that the window size is returned to its minimum value.
Referring now to FIG. 7, a message flow diagram for a transparent tree routing system is shown. As with the source routing system, the FR network issues a congestion notification with the congested DLCI to the FR Protocol element in the router R3. The FR Protocol element parses the frame so that the congested DLCI is in the form of a MAC address so that the LAN stations may understand it. As was mentioned above, each DLCI has a corresponding MAC address in a look-up table. The FR Protocol element maps the congested DLCI to its corresponding MAC address(es). The FR Protocol element forwards this information to its Station Manager in the form of a congestion notification along with the MAC addresses associated with the affected DLCI. The R3 Station Manager forwards this information directly to the LAN Manager which, in turn, sends it to the remaining Station Managers in its domain. As before, the Station Managers forward the information to their respective LLC Protocol elements which examine the affected MAC addresses against their outstanding LLCs. If there is a match, the Station Manager with the affected LLC will slow down its data traffic.
Thus, it can be seen that the method and system of the present invention provides a mechanism for efficiently interconnecting LANs through a frame relay network by allowing the LAN workstations connected to the LANs to provide flow control for data that the workstations are transmitting over the frame relay network when the network becomes congested. While the invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various other changes in form and detail may be made without departing from the spirit and scope of the invention.